This invention relates to a method for correction of thin film growth temperature and, more particularly, a technique for accurately controlling actual temperature of a semiconductor substrate during the process.
The technique for accurately measuring the actual temperature of a semiconductor substrate (hereinafter, referred to as actual temperature of substrate) during the process is important in various semiconductor processes. In particular, when thin film growth is performed with CVD (Chemical Vapor Deposition) equipment, high-precision temperature control is demanded because the actual temperature of the substrate largely affects such characteristics as the thickness of a deposited thin film, the uniformity and repeatability of resistivity, and the diffusion profile of a buried impurity diffusion layer.
As the method for measuring the semiconductor substrate temperature, there have conventionally been known those using thermocouples or optical pyrometers.
The thermocouple, in usual use, is buried in the center of the rear surface of a susceptor on which a semiconductor substrate is to be placed within semiconductor manufacturing equipment, in which case the semiconductor substrate temperature is determined based on the correlation between output of the heating source and the output voltage of the thermocouple. The thermocouple has been widely used by virtue of its large measurable range and good linearity of the correlation in measurement.
However, there arises a difference between temperature information derived from the thermocouple and the actual temperature of the substrate. This difference, indeed basically due to the fact that the target of measurement by the thermocouple is not the semiconductor substrate but the susceptor on which the semiconductor substrate is placed, but may arise within the same equipment unit, or between different equipment units, because of the setting of output of the heating source, the installation location of the susceptor or the replacement of parts. This would cause the actual process to progress at temperatures different from the set temperature.
On the other hand, the optical pyrometer, which is to determine the temperature of a heated semiconductor substrate by comparing the brightness of radiated light from the heated semiconductor substrate with the brightness of a standard lamp, is suitable for measurement in high temperature regions over 800.degree. C. However, with the optical pyrometer used, because the brightness is measured beyond the wall of a reactor made of quartz in the CVD equipment, the quantity of radiated light absorbed by the reactor wall does not hold constant due to the degree of contamination or thickness of the reactor wall. This makes it difficult to accurately measure the actual temperature of the substrate, disadvantageously.
As a means for solving these problems, there has been proposed a substrate-surface temperature measuring method using ion-implanted wafers in Japanese Patent Laid-Open Publication HEI 3-142948.
In this method, first, a plurality of diffusion wafers each having at the surface an impurity implantation layer of a specified concentration formed by ion implantation are prepared. These diffusion wafers are placed in a heat treatment furnace whose temperature characteristics are known, and subjected to heat treatment for a specified time under a plurality of different temperature conditions, so that ion-implanted impurities are diffused. Subsequently, by measuring sheet resistance of each diffusion wafer, a calibration curve representing the correlation between sheet resistances and known temperatures is prepared. Next, the diffusion wafers are placed in a reactor, which is the measurement target, and subjected to heat treatment at a thin film growth temperature for the same time duration as the heat treatment time in the preparation of the calibration curve. After this, sheet resistance of the diffusion wafers is measured by, for example, the four-point probe method. By determining the temperature corresponding to this sheet resistance value from the foregoing calibration curve, the actual temperature of the substrate placed in the measurement-target reactor can be known accurately.
However, in the technique disclosed in Japanese Patent Laid-Open Publication HEI 3-142948, at least one diffusion wafer made by ion implantation is needed each time the actual temperature of the substrate is measured. This diffusion wafer is unfortunately time-consuming and expensive, and moreover poorly available for 200 mm or more large-diameter types. Still, this diffusion wafer is made only for the measurement, and useless otherwise. Besides, even only the heat treatment for the measurement of the actual temperature of the substrate requires about one hour, to which the time required for the cooling of the diffusion wafer and the measurement of sheet resistance is added, so that at least two hours or more are spent for onetime measurement.
As shown above, the method using diffusion wafers, although superior in precision, may at a large possibility impair the economy and productivity when usually repeated.
Therefore, an object of the invention is to provide a method capable of solving these problems, and measuring and correcting the actual temperature of the semiconductor substrate at low cost in short time.